Parrot 3.0.0 Released | parrot.org | Log: irclog.perlgeek.de/parrot/today | Goals: Fix ipv6-related failures
Set by moderator on 10 February 2011.
whiteknight no 00:01
I mean, we could make it work 00:02
sorear the GC really ought to exist partly as macros
whiteknight requires some infrastructure additions
Hackbinary do people use perldoc to view the documentation? 00:03
arnsholt Yeah, I do that occasionally
lucian i see 00:04
Hackbinary do you use the linking between the documents?
arnsholt You can do that with perldoc? =) 00:05
lucian is probably procrastinating too much about C 00:06
Hackbinary is there away to only make something appear in perldoc rendering but not others? 00:07
dukeleto Hackbinary: define "others" 00:08
Hackbinary like html
what I'm trying to get to work properly is interlinking between the pod documents, and I can make an HTML only section, but if people are using perldoc, then it won't show up 00:09
maybe text?
00:13 ligne joined 00:15 GodFather left 00:20 bluescreen left
dalek rrot: 8128f59 | mikehh++ | MANIFEST:
re-generate MANIFEST
 00:29
rrot: c325eed | mikehh++ | tools/dev/merge_pull_request.pl:
fix codetest failure - cuddled else
rrot: b33f25e | mikehh++ | tools/dev/merge_pull_request.pl:
fix perlcritic failure - add perl coda
00:35 bluescreen joined
dalek rrot: 8ae0270 | cotto++ | docs/project/support_policy.pod:
clarify that Parrot changes which require code to be added to HLLs also consititude a breaking change
 00:37
dukeleto Hackbinary: i am not sure what the best solution is 01:06
Tene arnsholt: PAST's error handlers are certainly not optimal; I've had several ideas about improving them. 01:09
01:13 contingencyplan left
arnsholt My own semi-fluency in PIR and PAST doesn't help either =) 01:14
Tene arnsholt: what are you working on?
arnsholt Error handling in Parrotlog 01:15
Tene Nice.
arnsholt Somewhat unfortunately, the error recovery process is actually by far the most complicated bit =)
But I think the refactor (or perhaps reimplementation, more properly) will reduce the size of the code. Another happy side effect should be that it's actually correct, as well =) 01:16
01:19 dmalcolm left
cotto correctness schmorecctness 01:20
seen bacek 01:22
clunker3_ bacek was last seen on #parrot 10 hours, 47 minutes and 24 seconds ago, saying: night humans
aloha bacek was last seen in #perl6 9 hours 59 mins ago joining the channel.
cotto seen bacek
clunker3_ bacek was last seen on #parrot 10 hours, 48 minutes and 17 seconds ago, saying: night humans
aloha bacek was last seen in #perl6 10 hours ago joining the channel.
dukeleto updates wiki pages that talk about DEPRECATED.pod 01:24
arnsholt Whee. Looks like it's done for now. Better, and now it's only 19% of the source as opposed to 25% before 01:25
Ah, bollocks. Still bugged 01:29
A stack of exception handlers and time traveling continuations makes bookkeeping a tad tricky 01:30
dalek tracwiki: v30 | dukeleto++ | ParrotDeprecations 01:31
tracwiki: trac.parrot.org/parrot/wiki/ParrotD...ction=diff
tracwiki: v186 | dukeleto++ | WikiStart
tracwiki: trac.parrot.org/parrot/wiki/WikiSta...ction=diff
tracwiki: v187 | dukeleto++ | WikiStart
tracwiki: trac.parrot.org/parrot/wiki/WikiSta...ction=diff
tracwiki: v188 | dukeleto++ | WikiStart
tracwiki: trac.parrot.org/parrot/wiki/WikiSta...ction=diff
tracwiki: v8 | dukeleto++ | HowToDeprecate
tracwiki: trac.parrot.org/parrot/wiki/HowToDe...ction=diff
TT #1751 closed by dukeleto++: Links to DEPRECATED.pod et al
TT #1751: trac.parrot.org/parrot/ticket/1751
Hackbinary hi dukeleto, I have updated the PDD and run core sections with relative links that work for make html, a full url for github) 01:34
github.com/hackbinary/parrot/blob/...ossary.pod
arnsholt Oh, I know!
Hackbinary dukeleto, could you check to make sure you're happy with it, then I'll put through a pull request
arnsholt This requires pondering.
01:57 lucian_ joined 01:58 nwellnhof left, lucian left
mikehh oops forgot to report: 01:59
All tests PASS (pre/post-config, make corevm/make coretest, smoke (#9141) fulltest) at 3_0_0-918-g8ae0270 - Ubuntu 10.10 i386 (g++-4.5)
bacek ~~ 02:05
mikehh hi bacek 02:06
bacek aloha, mikehh
mikehh bacek: the powers that be decided it was too early to merge generational_gc branch, and dukeleto reverted the merge 02:07
bacek that's fine. 02:08
We can merge it back
after 3.3
According to deprecation policy.
mikehh bacek: it needs a deprecation notice apparently
bacek: I recon we need to merge it after 3.1 02:09
bacek mikehh, no. "Policy" now is pretty clear about it. 3.3 is earliest version.
mikehh bacek: bah I want it now, we need to discuss this in #ps 02:10
kid51 bacek: The good news is ... Parrot once again builds on Darwin/PPC with commit 77b57cc 02:11
bacek mikehh, yes, fine. I will probably miss next #ps. Got my riding test.
kid51, 77b57cc is what I mentioned in ticket?
kid51, or it's current master?
kid51 77b57cc is what you mentioned here: trac.parrot.org/parrot/ticket/2005#comment:2 02:13
(It also passes at current master HEAD, but of course that represents "pre-generational_gc" code. 02:14
bacek kid51, than it's good news. 02:15
cotto bacek, the current thinking is that it's worth making an exception to the deprecation policy
bacek cotto, I don't think so. There is no point to have deprecation policy with exceptions.
whiteknight blah 02:16
our current deprecation policy is broken
mikehh whiteknight: I totally agree with that 02:17
dukeleto Hackbinary: looking now
cotto we have commit bits. How do we fix it?
kid51 bacek: OTOH, it seems very slow; I started 'make' at 8:38 PM here; we're only halfway thru make test
whiteknight our users have been complaining about poor GC performance for years now, and we're saying that we have a solution but we can't deliver it out of respect for our users?
broken and stupid
bacek kid51, is it unoptimized build? Than it's doing a _lot_ of internal checks in GC. 02:18
dukeleto Hackbinary: you have broken links, such as github.com/hackbinary/parrot/blob/...d.pod.html 02:19
We need some good benchmark data for the new GC.
Who wants to make it happen?
kid51 bacek: Yes, this run is without --optimize 02:20
dukeleto the md5 benchmark is a good start, but we need more
bacek dukeleto, my benchmarks was "real life programs from our 'main customer'".
which is 25-40% faster with GenGC.
dukeleto bacek: you mean building rakudo ?
bacek All synthetic benchmarks was slower.
dukeleto bacek: yes, but that is a big macrobenchmark that is averaging stuff together
bacek dukeleto, "ops2c --core", "buiding rakudo" and "make specest" 02:21
dukeleto bacek: we have uneven performance effects here. We need to know which things get faster and which get slower and whether it is worth it
bacek dukeleto, lets use GC IMS. It's fastest implementation evar.
Hackbinary dukeleto, the html links won't work on github, because github is converting the pod files on-the-fly
so those will never work, thus the need for the separate 'github' links 02:22
dukeleto Hackbinary: email parrot-dev and see what people say
Hackbinary ok
will do so in the morning, it's 230 am here
=)
mikehh yeah I just noticed 02:23
Hackbinary :S
dukeleto bacek: i think we need some nqp-based benchmarks 02:25
bacek dukeleto, "ops2c --core" is nqp-based benchmark
dukeleto bacek: touche
bacek 40% improvements
mikehh but I have to do some work on a server in Southern California and can only really get access after 6pm there, for the resty of the weekend :-} 02:26
dukeleto bacek: does it have long-lived intergenerational PMCs?
mikehh bacek: BTW what is your "riding test" 02:29
kid51 t/compilers/opsc/02-parse-all-ops.t took 11 minutes to run
bacek dukeleto, a lot of them. 02:30
mikehh kid51: yup it seems to be the last test to finish for me in make test TEST_JOBS=4 02:31
bacek kid51, don't test performance on non-optimized builds. They a _really_ slow.
kid51 bacek: This may actually be an optimized build. But I won't be able to double-check until that terminal is free! 02:32
bacek kid51, :)
kid51 In any event, on this small-resource box, the build speed difference between '' and ''--optimize'' never leapt out at me. 02:33
cotto kid51, you can always look at what CONFIG_ARGS is set to in your Makefile
kid51 cotto: D'oh. Who was the guy added that to the Makefile? Me.
dukeleto running benchmarks on a desktop machine with lots of other programs running is also a bad idea 02:34
you need to run benchmarks at least 10x each and preferrably on a multi-CPU machine that is idle
kid51 Alright, so this is a non-optimized build. However, in master I usually expect make and make test to take on the order of 20-25 minutes. 02:35
dukeleto Tool::Bench can do the hard work: github.com/notbenh/tool_bench
cotto kid51, lawl
kid51 I'm now at 57 minutes and only just now got past t/compilers/opsc/06-opsfile.t
OTOH, I know that in master, I've been getting a test failure in t/op/number.t on the --optimize build on this box 02:36
This is akin to the big slowdown we had around mid-October of last year, when I was told, "It has to get worse before it gets better." 02:37
lucian_ is starting to like cyclone 02:38
kid51 Well, I'm not really trying to run benchmarks.
cotto I didn't like that game. kid51, thanks for making it known that the branch is slow on your box. 02:39
kid51 I'm trying to run make and make test in exactly the same way I've been doing since December 2006.
02:40 lucian_ is now known as lucian
kid51 dukeleto++ for closing TT #1751 02:46
02:47 bluescreen left
dukeleto de nada 02:47
02:50 whiteknight left 02:52 theory left 02:57 TonyC left 02:58 nopaste left 03:06 TonyC joined
arnsholt Damn. I'm not allowed to dynamically set the lexical context of a Sub. That's just mean :p 03:06
03:06 nopaste joined
dalek TT #2004 reopened by doughera++: Where did Configure.pl option '--cxx' go to? 03:08
TT #2004: trac.parrot.org/parrot/ticket/2004
arnsholt Tene: I think there's no sane fix for my problem without a patch to PCT to stop pasttype<try> from putting in pop_eh instructions. Could review/commit that? 03:11
Tene arnsholt: Only if you also include a fix for all other pct users... 03:22
arnsholt: you could could make a different pasttype, or stuff some metadata in the try node 03:23
arnsholt: within those constraints, you should have pmichaud review your proposal/patch 03:24
arnsholt My idea was a flag, like :isdecl for PAST::Var, that suppressed all pop_eh instructions 03:28
Not very sophisticated, but lets the user do whatever he wants with the exception handlers 03:29
Tene hmm... 03:30
I'll review a patch for you, sure
arnsholt Thanks! 03:31
03:36 ryan joined
Tene posting it anywhere? 03:37
arnsholt Haven't thought that far yet. Gotta make sure the code works first =) 03:44
But I guess git-format-patch and post to Trac? Or I could make a fork on GitHub and send pull request if that's better 03:45
dukeleto ~~ 03:49
Tene arnsholt: ah; ping me on irc or email me or whatever when you get it together. 03:50
I think I'm done for the evening. 03:51
mikehh All tests PASS (pre/post-config, make corevm/make coretest, smoke (#9156) fulltest) at 3_0_0-918-g8ae0270 - Ubuntu 10.10 i386 (g++-4.5 with --optimize)
Tene arnsholt: however you like. :)
dukeleto just created a smolder project for Cardinal 03:53
dalek rdinal: 0fb55be | dukeleto++ | / (2 files):
Submit smoke reports to smolder.parrot.org
dukeleto and now it should submit reports to the right place
Tene: ^^^
Hackbinary cool
Tene dukeleto: Thanks!
arnsholt Right. My minimal NQP test script gives the right results at least. Which testsuites should I run to make sure I didn't break something completely different? 03:55
Tene arnsholt: rakudo, at least.
Hackbinary I don't have access to edit trac, but could someone update trac.parrot.org/parrot/wiki/Languages with updated smoke reports url? 03:56
dalek TT #2004 closed by doughera++: Where did Configure.pl option '--cxx' go to?
TT #2004: trac.parrot.org/parrot/ticket/2004
dukeleto Hackbinary: what is your trac username ?
Hackbinary hackbinary ;) 03:57
dukeleto Hackbinary: ok, i will give you wiki edit perms. We had to lock it down due to spam.
Hackbinary yeah, I believe it ... damn spiced-hammers ;)
03:58 lateau joined
cotto Thanks for bearing with the inconvenience. 04:01
Hackbinary who me? no bothe 04:03
r
right ... I gotta crash .... it's way later than I meant it to be 04:04
talk to you tomorrow
dukeleto cotto: i can't seem to give hackbinary WIKI_EDIT in trac 04:05
cotto: can you figure it out?
cotto: i hate dealing with permissions on trac
cotto: i don't have the ability to give anybody any WIKI_* perms anymore, it seems 04:06
cotto dukeleto, sure 04:07
msg Hackbinary You should have a shiny new edit bit on the wiki. Don't use it to spam unless you really feel like it. 04:09
aloha OK. I'll deliver the message.
cotto dukeleto, do you want a trac admin bit? 04:11
dukeleto cotto: i used to have one
cotto: i don't know what happen to it
cotto: sure, i guess
cotto you seem to have one now
dukeleto cotto+_ 04:12
dukeleto facepalms
plobsing cotto+*
dukeleto i am submitting a failing smoke report for cardinal nowishly
cotto plobsing, whatever
04:26 kid51 left
mikehh rakudo (d267ba8) - builds on parrot (3_0_0-918-g8ae0270) - make test, make spectest_smolder[(#9181), roast (b3ef08d)] PASS - Ubuntu 10.10 i386 (g++-4.5 with --optimize) 04:47
27,634 ok, 0 failed, 610 todo, 1,847 skipped and 0 unexpectedly succeeded
Tene dukeleto: I get a 404 when running rake smolder 05:08
for smolder.parrot.org/app/public_proje..._report/10
05:10 jsut joined 05:15 jsut_ left
cotto dukeleto, ping 05:17
github++ 05:22
mikehh docs/project/committer_guide needs updating 05:36
cotto yes it does 05:37
It'll be a joy to delete much of that.
mikehh, done 05:44
dalek rrot: 6d69dec | cotto++ | docs/project/committer_guide.pod:
delete and update some obsolete svn workflow docs
cotto mikehh++
there are still a depressing number of references to svn in the codebase 05:45
05:47 ryan left
mikehh cotto: trying to add the rest of docs/project to make html (just checkin' the results) [local branch atm] 05:55
cotto good idea
mikehh we need to see what other docs need to be added 05:56
arnsholt Tene: Got a couple of Rakudo spectest failures with my patch. Rerunning it now without the patch to make sure they don't go away 06:05
06:07 rurban_ joined 06:10 rurban left, rurban_ is now known as rurban
arnsholt Tene: TT #2008 06:56
dalek TT #2008 created by arnsholt++: [PCT] Patch/feature request: No pop_eh in try PAST::Op 07:09
TT #2008: trac.parrot.org/parrot/ticket/2008
07:17 fperrad joined
Tene arnsholt: I'll read it eventually, but I recommend you also contact pmichaud to ask for comments on this design. 07:49
cotto +1. pmichaud wrote that code and knows it best. Unfortunately he's on vacation for the next 5 (iirc) days, so it'll be a while before he'll get back to you. 07:55
Tene Ahh. 08:42
I'll review it for sanity tomorrow, then, so i can at least give early feedback.
09:03 protorom joined
lateau Tene: broken URL modified and I want to send my patch 09:58
Tene: sry. it is about Cardinal rake smolder task.
Tene: I sent a patch to parrot-dev and check it when you have blank time. 10:10
11:09 jsut_ joined 11:13 jsut left 11:15 whiteknight joined 12:00 contingencyplan joined 12:03 ligne left 12:10 jsut joined 12:14 jsut_ left 12:35 contingencyplan left 12:53 eternaleye left, slavorgn left, wagle_ joined 12:54 eternaleye_ joined, snarkyboojum_ joined, nopaste left, lucian left, AzureStone left, simcop2387 left, Hunger left, wagle left, cognominal left, KaeseEs left, Hackbinary left, Maddingue left, aloha left, jnthn left, he left, dukeleto left, [hudnix] joined, eternaleye_ left, whiteknight left, protorom left, mj41 left, sri left, mikehh left, particle1 left, cotto_work left, perlite left, Kulag left, autark left, arnsholt left, ttbot left, hatseflats left, frodwith left, jasonmay left, Infinoid left, Util left, confound left, elmex left, rblackwe_ left, [hudnix] left, snarkyboojum_ left, jsut left, rurban left, TonyC left, plobsing left, Patterner left, PerlJam left, Coke left, jan left, preflex left, allison left, silug left, dngor left, dalek left, sorear left, szabgab left, jjore left, pjcj left, fperrad left, lateau left, hudnix left, clunker3_ left, NotFound left, GeJ left, PacoLinux left, Tene left, snarkyboojum left, cxreg left, he_ left, TiMBuS left, bacek_at_work left, luben left, pmichaud left, p6eval left, atrodo left, zostay left, moritz left, szbalint left 12:55 atrodo joined 12:57 [hudnix] joined, snarkyboojum joined, eternaleye_ joined, jsut joined, whiteknight joined, protorom joined, fperrad joined, rurban joined, lateau joined, TonyC joined, plobsing joined, mj41 joined, sri joined, Patterner joined, PerlJam joined, mikehh joined, clunker3_ joined, particle1 joined, Coke joined, jan joined, cotto_work joined, preflex joined, perlite joined, Kulag joined, autark joined, allison joined, GeJ joined, arnsholt joined, silug joined, ttbot joined, dngor joined, frodwith joined, hatseflats joined, rblackwe_ joined, Util joined, confound joined, jasonmay joined, elmex joined, Infinoid joined, Tene joined, dalek joined, sorear joined, jjore joined, szabgab joined, pjcj joined, cxreg joined, he_ joined, TiMBuS joined, bacek_at_work joined, luben joined, pmichaud joined, p6eval joined, zostay joined, moritz joined, szbalint joined, PacoLinux joined, Hunger joined, Themeruta joined, Maddingue joined, AzureStone joined, nopaste joined, lucian joined, simcop2387 joined, cognominal joined, KaeseEs joined, Hackbinary joined, aloha joined, he joined, dukeleto joined, jnthn joined 13:00 slavorgn joined 13:21 protorom left
tadzik gist.github.com/823767 -- could someone git-strong give me a hand? 13:44
13:52 kid51 joined
plobsing tadzik: git checkout -b gen_gc2 origin/gen_gc2 13:52
I think that's what you want
dalek nxed: r805 | NotFound++ | trunk/winxedst (2 files):
new predefs getstdin, getstdout and getstderr
 14:02
14:02 kid51 left 14:05 ambs joined 14:07 rurban_ joined 14:10 rurban left, rurban_ is now known as rurban
dalek nxed: r806 | NotFound++ | trunk/winxedst (2 files):
fix mistakes from r805
 14:13
tadzik plobsing: thanks, that was it 14:14
14:16 [hudnix] left 14:18 ambs left 14:19 ambs joined
lucian slavorg: awesome name 14:19
14:28 vmspb joined 14:51 lucian_ joined 14:54 lucian left
dalek tracwiki: v149 | hackbinary++ | Languages 14:56
tracwiki: Updated cardinal smoke reports link
tracwiki: trac.parrot.org/parrot/wiki/Languag...ction=diff
15:23 kid51 joined
kid51 afk 15:30
15:31 hudnix joined
dalek p: c8ce669 | jnthn++ | src/pmc/nqpmultisig.pmc:
Fix a mark bug that saw some people get segfaults in multi-method dispatch situations.
 15:49
p: 5ed3fad | jnthn++ | src/metamodel/reprs/P6str.c:
Fix an issue in the P6str repr spotted by sorear++.
 16:00
16:03 dngor left 16:05 vmspb left 16:16 cognominal left 16:17 Psyche^ joined, ryan joined, Patterner left, Psyche^ is now known as Patterner 16:37 cognominal joined 16:38 dngor joined, whiteknight left 16:54 plobsing left
Coke folks may be happy to know that the segfaults I was seeing appear to be gone in master. 17:09
17:13 plobsing joined
cotto I'm happy. 17:24
dalek rrot: 6742527 | mikehh++ | docs/index/developer.json:
add remaining docs/project pod files to new section in developer docs
 17:39
rrot: 9dd88ad | mikehh++ | docs/project/core_inclusion.pod:
fix doc so title is correct
rrot: dc55b45 | mikehh++ | docs/project/committer_guide.pod:
merge in latest master
rrot: 63da9c3 | mikehh++ | docs/project/committer_guide.pod:
change a couple of remaining refs to svn to git
rrot: d76e2cc | mikehh++ | docs/project/release_manager_guide.pod:
Change first heading so title is Release Manager Guide
17:39 jan left
kid51 mikehh: Have you been able to take another look at the 'make html' problem? trac.parrot.org/parrot/ticket/1954 17:43
17:44 jan joined
mikehh kid51: a look yes, need to spend some time on it after the release 17:45
kid51 Okay. It's non-urgent.
mikehh kid51: it is also tied up with my failed attempts on html_cleanup_2 (every time I tried to remove something, got failures - got to sort out dependencies) 17:47
kid51 Hmm. I wasn't aware of that branch (or perhaps I've got too much gen_gc on my mind). What is its objective? 17:48
mikehh cotto: I never got proper access to SSH into ftp-osl.osuosl.org, I have changed my keys since then as well
kid51: it was local, I never pushed it upstream as I never got it to work properly 17:49
17:51 theory joined
mikehh kid51: the objective was to remove all the stuff from the old make html especially lib/Parrot/Docs/Section/*.pm etc 17:56
kid51 Aha, I remember you mentioning that.
cotto mikehh, let's fix that now.
kid51 cotto: provided 'now' means 'after our release' 17:57
cotto ?
mikehh how can I get the ./ssh/id_s 17:58
dsa_pub key to you
kid51 cotto: When you get a chance, can you take a glance at trac.parrot.org/parrot/ticket/540?
cotto mikehh, nopasting is fine. The only thing it lets someone do is give you access to their machine.
kid51, deal
kid51 Issue of long-standing, non-urgent, but two key developers w/different opinions 17:59
cotto That's a long ticket there.
kid51 long as well as long-standing, then
cotto grabs a shovel and digs in
mikehh, as long as the file has "pub" in the name, putting it on the internets is fine. 18:01
anyway, I'm ready whenever you are. 18:02
nopaste "mikehh" at 192.168.1.3 pasted "~/.ssh/id_dsa_pub" (1 line) at nopaste.snit.ch/31480 18:03
mikehh cotto: one of the problems I have is that I have separate partitions for various versions of Ubuntu/Kubuntu with different desktop id's 18:05
cotto mikehh, is mhb-desktop your old machine?
mikehh, you can copy the same public/private key between them or add different keys for the same machine. 18:06
mikehh cotto: yeah I think it was Ubuntu 9.10 amd64,
cotto ok. I'll delete that one. 18:07
mikehh, you should be able to log in to that machine now
mikehh cotto: what would I use for login? 18:08
cotto use the username parrot
mikehh cotto: 18:09
18:09 contingencyplan joined
mikehh bah hitting the wrong key again, enter instead of backspace 18:09
cotto I wonder if a ccache equivalent for parrot would make a viable gsoc project. 18:12
kid51 "viable" should include "completable in the GSOC timeframe" (IMO) 18:14
cotto of couse 18:15
that does get into the problem of cache invalidation, so I suspect it's a bit deeper than one might expect 18:16
mikehh, can you get to the ftp site? 18:17
mikehh cotto: seems to work for now :-}
cotto excellent 18:18
do you have access to the vm to update docs.parrot.org? 18:19
18:23 lateau left
mikehh cotto: I can log in to www.parrot.org as Mikehh but I don't have admin access 18:23
cotto: mikehh but it displays Mikehh 18:25
nopaste "kid51" at 192.168.1.3 pasted "t/compilers/imcc/syn/errors.t: Make test pass make warnings_tests: Okay to commit?" (13 lines) at nopaste.snit.ch/31481 18:28
cotto ok
@mikehh
mikehh, I don't have an admin bit on parrot.org 18:29
I think the capitalization is just drupal being wacky. 18:30
dukeleto, ping 18:50
18:51 ambs left
dukeleto cotto: pong 18:54
cotto dukeleto, I'd appreciate your thoughts on gist.github.com/88eb00a0c2296eeb48e2 , my idea on how to build and refine the M0 spec as a team. 18:55
kid51 text wrap is a desirable thing 18:59
cotto yeah 19:00
let me fix that
done
kid51 new PDD++ 19:02
cotto it's in the m0-spec branch
kid51 thinking about how a team will function++
mikehh: One small patch to a test for your pre-release consideration: trac.parrot.org/parrot/attachment/t...rrors.diff 19:03
cotto I'd be further ahead at this point if I just did it myself, but hopefully an effective way of getting more brains involved will result in a better final spec.
kid51 cotto: Are you using gist in preference to your own blog? Perhaps something on your blog about this so that planet.parrotcode.org will pick it up. 19:04
dukeleto cotto: i likes it 19:05
cotto kid51, I'm using it to draft an email that I can also easily share with others before sending it out.
dukeleto cotto: if you could add that we are looking for ideas that are implementable in our first prototype, and not so much "pie in the sky" ideas, that would be useful, i think
cotto: i want to put in peoples minds that open-ended opinion-discussing is not as useful as "let's try this for our first prototype" 19:06
kid51 dukeleto++ We have to dispel the feeling that Lorito is vaporware
dukeleto cotto: and make sure people know which features we know will not be part of the first prototype
kid51: yes, exactly 19:07
kid51 afk
cotto dukeleto, good suggestions
dukeleto cotto: i want volunteers that want to help make the 1st prototype happen, not those that want to soak up lots of time debating about the color of the bikeshed 19:10
cotto dukeleto, I consider it an important part of my job to say "The bikeshed will be green. Next topic." 19:11
19:11 rantanplan joined
cotto dukeleto, I'm going to be making a conscious effort to avoid or short-circuit those kinds of questions. 19:12
updated 19:15
19:17 lucian joined, rantanplan left
dukeleto cotto: awesome 19:17
mikehh kid51: testing it now 19:19
19:20 lucian_ left
cotto I'll sit on it an hour, re-read and send it out then. 19:20
19:25 kid51 left
dalek rrot: b058323 | mikehh++ | t/compilers/imcc/syn/errors.t:
incorporate a patch suggested by kid51++
 19:33
rrot: d9d7674 | dukeleto++ | docs/project/metacommitter_guide.pod:
[doc] Update the metacommitter guide for a Git New World
rrot: 8c179b3 | dukeleto++ | lib/Parrot/Distribution.pm:
[cage] Remove dead code from Parrot::Distribution
 19:49
rrot: 3b6bed1 | dukeleto++ | lib/Parrot/Harness/Smoke.pm:
[docs] Clean up Parrot::Harness::Smoke docs and s/svn/git/
 19:54
dukeleto cotto: that is about as much s/svn/git/ cleaning as I am up for doing
cotto: it looks like we have lots of dead perl code relating to svn
cotto I saw a bit last night but got distracted. 19:55
dukeleto, I think it's what you nuked just now.
dukeleto cotto: nope, there is more 19:58
cotto: i just made a TT
dukeleto just learned about "git branch --no-merged" 19:59
which will list local branches that are not reachable from HEAD
so it shows unmerged branches 20:00
pretty useful
dalek TT #2009 created by dukeleto++: Remove dead svn-related code from Perl modules 20:02
TT #2009: trac.parrot.org/parrot/ticket/2009
cotto We also need a ticket for the write barrier
I'm not sure why I said that instead of just doing it. 20:03
dukeleto cotto: do it :) 20:05
i just aliased "git unmerged" to "git branch --no-merged" and life is a little better.
cotto is doing it 20:06
bacek, ping 20:07
mikehh All tests PASS (pre/post-config, make corevm/make coretest, smoke (#9303) fulltest) at 3_0_0-926-gd9d7674 - Ubuntu 10.10 i386 (g++-4.5 with --optimize) 20:09
cotto bacek, unping
msg bacek How beneficial might it be to add an :ro attr that prevents pmc2c from generating write barriers for a C-level METHOD? 20:12
aloha OK. I'll deliver the message.
cotto That's a tricky ticket to describe. 20:27
dalek TT #2010 created by cotto++: add write barrier to all PMCs that write in unusual VTABLE functions 20:35
TT #2010: trac.parrot.org/parrot/ticket/2010
rrot: f1c998b | cotto++ | api.yaml:
add proper notice for the write barrier change
20:37 zby_home joined 20:39 fperrad left
cotto atrodo, ping 20:41
atrodo, unping 20:43
20:47 bacek_mobile joined
bacek_mobile ~~ 20:47
cotto, We do support :write for METHODs. Unfortunately it's not used consistently in (at least) our PMCs.
I'm pretty sure, that HLLs aren't use :write properly. 20:48
cotto bacek_mobile, I know a guy with a commit bit.
bacek_mobile cotto, it will require big review of our pmcs first
I didn't want to increase scope of gen_gc branch to such amount of work. 20:49
cotto Where'd those gci minions go.
bacek_mobile, that's sensible.
bacek_mobile for gen_gc scope was: 1. Make it. 2. Make it correct. 20:50
cotto bacek_mobile, is you at you dot com your preferred email?
bacek_mobile cotto, yes :)
cotto, you can create ticket similar to #2010 for properly mark METHODs with :write pragma. 20:52
cotto bacek_mobile, do you think it's worthwhile? 20:54
bacek_mobile cotto, yes.
cotto I was thinking a :ro pragma would be easier.
Util dukeleto: ping
bacek_mobile cotto: :ro is same is !:write. 20:55
and we already have :write
cotto yes, but it's a question of blacklisting vs whitelisting
bacek_mobile check src/dynpmc/rotest
afaik, most of the methods are ro. 20:56
at least in our code.
cotto for some it depends on the arguments 20:57
bacek_mobile I'm not sure that pmc2c support multiple pragmas for methods. 20:58
and anyway, :ro isn't correct pragma for this purpose
cotto looks like it parses them, at lesat 20:59
bacek_mobile something like :immutable is better choice
cotto *least
I'll let the guy who commits pain that bikeshed.
*paint
Util cotto: I might like that version better that "paint". :) 21:00
mikehh rakudo (f4ec358) - builds on parrot (3_0_0-928-g3b6bed1) - make test, make spectest_smolder[(#9312), roast (051af06)] PASS - Ubuntu 10.10 i386 (g++-4.5 with --optimize) 21:17
27,634 ok, 0 failed, 610 todo, 1,841 skipped and 0 unexpectedly succeeded
Util is afk; wrestling with a fallen tree.
cotto Util, get it in a headlock. That's their secret weakness. 21:18
dalek rrot/m0-spec: 685db46 | cotto++ | docs/pdds/draft/pdd32_m0.pod:
add a section on objects
 21:21
rrot/m0-spec: 0b91df4 | cotto++ | docs/pdds/draft/pdd32_m0.pod:
add status and some milestones
mikehh winxed (906) - builds on parrot (3_0_0-928-g3b6bed1) - make test/test1/test2 PASS - some examples ok tried fly, opengl and others. 21:28
sorry that should be 806 21:29
cotto atrodo, ping 21:31
atrodo, unping 21:33
21:42 perlite_ joined 21:44 plobsing left 21:45 perlite left, perlite_ is now known as perlite 22:06 bacek_mobile left 22:07 rurban_ joined 22:10 rurban left, rurban_ is now known as rurban, plobsing joined
dalek rrot: f7d6c26 | cotto++ | src/pmc/packfileannotations.pmc:
remove a couple unused variables, fixing CID 1288
 22:14
22:30 plobsing left
dukeleto ooooh, the GCC Compile Farm just got a netbsd 5.1 machine 22:44
23:07 plobsing joined 23:19 TonyC left, nopaste left
dalek p: d6338b2 | jnthn++ | src/NQP/Actions.pm:
Toss 'is pirflags' and add 'is vtable'. Calls method on the meta-object, though none implement it yet. Eventually this allows overriding of some Parrot v-table methods again when using 6model, which we'll need for other Parrot bits to interop with languages using 6model.
 23:29
p: a022547 | jnthn++ | t/nqp/52-vtable.t:
Update test file to use is parrot_vtable rather than is pirflags. Still doesn't pass again yet, though.
23:46 lucian left 23:49 jsut_ joined 23:54 jsut left
dukeleto anybody know anything about NanoJIT? www.bluishcoder.co.nz/2009/05/simpl...ample.html 23:58
Nanojit is a small, cross-platform C++ library that emits machine code. Both the Tamarin JIT and the SpiderMonkey JIT (a.k.a. TraceMonkey) use Nanojit as their back end.
might be a good place to learn from for our JIT 23:59

Powered by Cro and the Raku® Programming Language.

Please report any issues / comments / feature requests as an issue on App::Raku::Log.

Thank you!